Higgsfield AI released a project called Hell Grind that the company and some outlets have marketed as the "first 95-minute AI movie," according to coverage by Notebookcheck and WorldOfReel. Reporting by The Wall Street Journal and other outlets says the San Francisco startup produced the film in roughly two weeks on a budget of $500,000, with 80% of that reportedly spent on compute. The production used ByteDance's Seedance 2.0 as part of the pipeline, according to Notebookcheck. The film screened at industry events in Cannes city, not as part of the official Festival de Cannes selection, per AV Club and WorldOfReel. Only the first roughly 22 minutes are publicly available so far, and early online reaction has been largely negative, with critics calling the visuals artificial and the story weak, according to Notebookcheck, AV Club, and WorldOfReel.
What happened
Higgsfield AI unveiled Hell Grind, a 95-minute action film it and several outlets have billed as the "first 95-minute AI movie," according to coverage by Notebookcheck and WorldOfReel. The Wall Street Journal reports the San Francisco startup assembled the project in about two weeks on a budget of $500,000, and that roughly 80% of the spend went to compute. Notebookcheck and other outlets report the production used ByteDance's Seedance 2.0 as part of the generative-video pipeline. The film was shown at market events in Cannes city but was not part of the Festival de Cannes official selection, per AV Club and WorldOfReel. As of reporting, only the first roughly 22 minutes (episode 1) are publicly available; Higgsfield is releasing additional segments episodically, Notebookcheck reports.
Technical details
The Wall Street Journal and AV Club report that Higgsfield relied on high-volume prompt-driven generation: the WSJ coverage cites roughly 16,181 initial video generations for the first 25 minutes, distilled into 253 final shots, with prompts described as averaging about 3,000 words each. Notebookcheck additionally identifies Seedance 2.0 in the toolchain. These sources frame the production as a heavy compute play where large-scale clip generation and manual curation produce a linear film-length sequence from many short renders.
Editorial analysis - technical context: For practitioners, this production illustrates two consistent patterns in current generative-video workflows. First, scaling to long-form content still depends on exhaustive generation plus human selection and editorial assembly rather than a single end-to-end model output. Second, compute and iteration costs remain the dominant line item; the WSJ figure that compute consumed most of a $500,000 budget is consistent with other recent large-scale generative-video experiments.
Context and significance
Coverage places Hell Grind at the intersection of product demonstration and market positioning. The WSJ reports Higgsfield intended the film as a showcase for studio buyers, while AV Club and WorldOfReel flagged the distinction between market screenings in Cannes city and inclusion in the festival program. Public critical reaction has been mixed to negative: Notebookcheck, AV Club, and WorldOfReel describe audience comments calling the visuals artificial, the editing choppy, and the narrative shallow. That reception underscores a broader debate reported at Cannes about how, and where, generative AI should be introduced into film production; the WSJ quotes actress Demi Moore saying, "AI is here. And so to fight it is to fight something that is a battle that we will lose."
For practitioners: The project is notable as a demonstration of capability rather than a consensus artistic success. It documents workflow practices-mass generation, heavy compute, manual shot curation-and provides a concrete cost and time benchmark for teams exploring long-form generative video.
What to watch
- •Distribution and festival placement, including whether future screenings clarify official selection status or move into mainstream festivals. This is a public indicator of industry acceptance.
- •Studio interest and commercial licensing conversations reported by trade press, which would signal business traction for AI video platforms.
- •Technical metrics: reduction in per-shot generation counts, prompt lengths, or compute spend reported by vendors, which would show efficiency gains.
- •Legal and rights discussions, especially around actor likeness, union responses, and attribution, which have been raised in parallel coverage of AI in filmmaking.
Scoring Rationale #
The story provides a concrete, public benchmark for long-form generative-video production costs, workflows, and audience reception. It is notable for practitioners evaluating tooling and commercial use, but it does not introduce a new foundational model or change core research directions.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.